Skip to content

IAI Actuarial Core Principles · Risk Modelling and Survival Analysis · Elementary principles of machine learning

A claims-classification model is trained with early stopping. Training error keeps decreasing each iteration, while validation error decreases for 40 iterations and then starts rising. What is the sensible action?

Stop training at around iteration 40. After that point validation error rises even though training error keeps falling, which signals overfitting, so the lowest validation error identifies the model most likely to generalise to new data.

  1. AStop training at about iteration 40, where validation error is lowestCorrect
  2. BContinue training until training error reaches zero
  3. CStop training at iteration 1, where the model is simplest
  4. DUse the validation data as extra training data and continue iterating
  5. Ignore validation error because only training error measures fit

Explanation

Rising validation error after iteration 40 shows the model has begun to overfit while still improving on training data. Early stopping picks the iteration with the lowest validation error. Continuing until zero training error would deepen the overfitting.

Did you get it right without looking?

One question tells you little. A timed set on Elementary principles of machine learning shows your real accuracy, how long you take and where you lose marks.

Practise Elementary principles of machine learningFree diagnostic mock

More Elementary principles of machine learning questions